Rhaegal

Using that formula and those boss/tank weapon speeds (seems like a fair guess for the average fight to me), you're looking at ~8.6% increased damage done to the tank by the boss's autoattack (i.e., if the boss also hits the tank with special abilities, the total damage the tank takes over the fight will be increased by a smaller amount). Assuming an 11.2% cap, it would require 11.25 * 4 * 3.94 = 178 expertise rating to eliminate that extra damage. On fights where the boss's attack speed is significantly faster, the 8.6% would scale down proportionately, unless you're in an extreme circumstance (e.g., Prince phase 2) where the extra attacks can result in crushing blows.

If you're really looking for a "what is this stat worth in mitigation compared to dodge", for example, you could easily enough calculate off those formulas. I'm not going to because I still think that's a silly way to look at it because it's not a mitigation stat, it's a threat and insta-gib protection stat, and therefore only going to be useful on fights were those are an issue.

[edit: math error]

Stand back! I'm going to try SCIENCE!

 

Rhaegal

So I kept thinking about this and decided that just for the hell of it, I'd actually make a real comparison between expertise and dodge. DISCLAIMER: Assuming my math is right, this is only true for the specific warrior from my guild that I armoried, using a 1.6 speed weapon, 16.5% dodge, 16.3% parry, 11.5% miss, and against a boss with a 2.5 speed attack. The actual relationship will vary slightly depending on a specific tank's avoidance, and will vary significantly more depending on the type of tank (warrior vs. bear vs. paladin), and the attack speed of the boss. Insert entire shaker full of salt grains. (And also remember that for the purpose of calculations, I'm ignoring that expertise rating is only useful in chunks of ~4.)

For every 1 expertise (3.94 rating) you have, the boss's parry chance gets reduced by 0.25%. To fully eliminate parry, and ultimately reduce the average haste from parries to 0, you'd need 11.2/0.25 = 45 expertise, or 178 expertise rating. Every point of expertise will give 8.6/45 = 0.19% reduced damage for a warrior (~0.05% per rating).

If you want to compare to dodge, it gets even more complicated. Dodge is hard to calculate, because the actual relative amount of reduced damage per 1% dodge depends on the tank's current avoidance. For example, 1% dodge for a hypothetical warrior with 0% miss, 0% dodge, and 0% parry reduces the total amount of damage he will take by 1%. On the other hand, if you have a warrior with 20% dodge, 20% parry, and 10% miss (50% total avoidance), adding 1% dodge to the table is twice as effective, reducing total damage taken by 2%. Let's take some generic warrior (by that, I mean our MT in mostly ZA and badge gear, with a few leftover Kara pieces) with, say, 16.5% dodge, 16.3% parry, and 11.5% miss, or 44.3% avoidance. 1% dodge gives him 1.8% damage reduction over time. At 18.92 rating per 1% dodge, that means dodge is worth 0.095% damage reduction per rating, roughly twice that of expertise. For this specific warrior, expertise rating yields the increased threat effect it's intended for, as well as HALF the amount of mitigation as dodge rating. HOWEVER, this "mitigation" from expertise ONLY effects the damage the boss will do with its autoattack, whereas if it has special physical abilities that can be dodged, dodge will pull far away from expertise in terms of mitigation.

Comparing it to defense rating would be obnoxiously complicated due to trying to factor in block. Armor comparisons are pretty easy but change significantly by individual, so can be calculated using the "~0.05% per rating" number and comparing to your specific amount of armor.

I really hope no one takes this as "SWEET, expertise gives half as much avoidance as dodge!" and runs with it. You still can't make a direct comparison without caveat after caveat, and adjusting it from fight to fight, which is the point I'm trying to make.

Stand back! I'm going to try SCIENCE!

 

ALEXTREBEK

Expertise is no substitute for defense or dodge, whatsoever. Expertise as a mitigation stat is in an entirely different category than defense or dodge. Like I said before, Expertise is quite a bit more like resilience than anything else in its ability to dampen burst damage. Just as resilience is key in PvP concerning the mitigation of burst damage, so too is expertise key in PvE concerning the mitigation of burst damage.

To put it simply: If a boss gains parry haste and subsequently attacks faster than it ordinarily would, avoidance will still reduce the attack's chance to hit you. Now, if you stack expertise, the boss's hastened attack will have a lesser chance of occurring. Similarly, when you stack resilience in PvP, it reduces the chance of an opponent's critical strike occurring. If you PvP you should know that Resilience is no substitute for actual Health (which allows you to take more burst damage), and you should realize that Expertise follows suit in that it should be no substitute for actual Health or avoidance.

Whether Expertise should be stacked over health or avoidance, well, I personally don't believe it should, but it certainly should not be overlooked; and, at this stage of the game, tanking T6 instances with any less than 20 Expertise is like running into the 2,000s bracket with like 175 Resilience.

 

Taliafears

I haven't calculated it all out yet, but I have a very rough idea on a way to approach this problem.

It occurs to me that hit-parry-parry-hit (I get hit once, the boss parries me twice, and his next hit comes really quickly and lands, hereafter referred to as a gib) approaches the severity of taking a critical hit. Obviously we'd like to make sure this doesn't happen. So, what are the chances of getting a gib?

For any given boss attack, the chance that this attack will start a gib sequence is 1-(hit^2*parry^2), where hit is the boss' chance to hit me, and parry is his chance to parry me.

So the incremental value of dodge can be found by taking the derivative:
dhit/dgib = -2*hit*parry^2
Likewise with expertise:
dparry/dgib = -2*hit^2*parry
And divide to compare them to one another:
dhit/-dparry = -parry/hit

so at 10% parry and 50% hit, -1% parry = 5% dodge
and at 5% parry and 50% hit, -1% parry = 10% dodge

It seems that we have increasing returns on the gib-prevention aspect of expertise.

Of course, in the end this is just skirting around the issue. The real issue is how important is gib-prevention? I'd really like to spend more time on the topic but that's my extremely rough initial stab at it.

 

mhr_78

If i hook into this discussion with a funny combat log parse from last nights raid, enjoy:

Thats 22k damage in less than 1.5 seconds from a BT boss, and i've actualls seen gorefiend crush way higher in older parses. Now the thing isnt really an gear issue but it's either the fight is a bit unbalanced when at any time the boss can practically one shot your tank.

Would capping expertise reduce the chance of these hasted crushing blows? As gorefiends attack table always has a chance of sneaking crushing blows through before the shieldblock CD is ready - i'd guess stamina, armor and expertise would be the best mitigation on ridiculous frontload damages for warrior tanks.

It's ridiculous though, i think azgalor has had crushing blows removed from his attack table just for that reason. And to be honest, i'd rather have crushing blows removed as well. It's retarded if the CPU rolls a dice and wipes you.

 

Thorongil

Basically, that´s what expertise is about - reducing those unlucky parry-streaks.

Apart from that I´d say that it is known that Teron hits quite hard thus 1,3 seconds without any kind of heal also means that there has to be some part of the guilt on the healers´ part. Fully stacked HoTs and GHs in the pipe could probably have prevented that death, nonetheless sometimes such spikes happen and can´t be countered due to various reasons.

By the way - just to pick up your spark - concerning Azgalor and his incapability of crushing: There is one major difference. Azgalor has a 5 second silence that can´t be countered reliably throughout the whole fight (by the healers). This is not the case with Teron. There is no reason why healers are forced to stop healing the tank (apart from being ghosted, of course) thus crushing blows do not seem very unfair or luck-based in that fight to me. The dice the CPU is rolling can be countered by dedicated healers most times. However, I am speaking from the position of fully BT-geared tanks, which might be a bit unfair considering how early you encounter Teron.

Farewell, O twice beloved! A Turin Turambar turun ambartanen: master of doom by doom mastered! O happy to be dead!

 

kalbear

I wrote a blog article on this. The conclusion I reached was that at all levels, expertise is far more effective than avoidance at reducing your chances of parry gib - twice as good at all reasonable levels of dodge and expertise, essentially. EH is good as well, but expertise quickly reduces that chance. That it also has a side benefit of threat is a great deal for stacking. This is true for any tank, but especially true for warriors and druids.

 

◊ Fellwraith

Was imp demo shout up (and if not, was CoR up)? How about thunderclap? Did your warrior use ironshields?

I wear hybrid dps gear for tanking Teron now and I don't see anywhere near that much damage on a crushing blow or a normal hit (10-11k is the max for me, I don't wear a sunwell tanking suit for this fight). It's not ridiculous if you are buffing/debuffing appropriately and have the right buffs on your tank. Expertise might not have saved your tank, but using everything available to him would have. Taking 18k in the space of 1.5 seconds is survivable. He took that much damage because he is either undergeared (ie wearing all T4/kara tanking gear), not buffed properly or not debuffing gorefiend properly.


Except you're neglecting the fact that dodge also protects you from the preceeding (non-parry flurried) hits and possibly saves shieldblock charges. Also dodge prevents special attacks from landing on you. If you were really comparing the two stats you would have to apply your dodge rate to all mob attacks in the series you're analyzing and you would need to factor the tank's swingspeed and special attacks into the equation. It's not a simple calculation.

They're both useful stats, overstacking one to the exclusion of the other for every fight and every class isn't a smart move. There's plenty of fights where dodge is more useful. Sometimes the spike comes from the auto-attack and a special attack. What good does reducing parry flurry do for you in that situation?

Also, I'm pretty sure the max swing speed increase is 40%, the average is about 25%. You're quoting the absolute worst possible parry chance and assuming that a parry will instantly be a gib, when that isn't always the case. There's something to be said for reducing your average damage taken in a fight.

 

kalbear

No, it's very simple. You only have to take into account one special case: when you do not avoid an attack followed by a parry-hasted attack. That's it. Shield block charges don't matter here; when you get parry gibbed is not when you get crushed, it's when you get multiple hits quickly. The only question you need to ask is what the situation where a tank can die due to burst damage is. I took the view that the only time this can happen without external environmental issues occurring (ie, healers silenced/feared/whatever) is when a tank takes two fast hits in a row. If you believe that it's 3 fast hits, that's modelable as well. You can create other situations where the equation is different (for example: a normal hit + special hit + hasted normal hit, and what the chances are of that). Heck, let's do that. The chances of a normal hit + special hit + hasted normal hit are:
(1-avoidance)*(1-avoidance)*(parry chance) * (1-avoidance). We'll assume a 15% parry chance as well.

For 50% avoidance, no expertise this is 1.875%
For 51% avoidance, no expertise this is 1.7647%
for 50% avoidance, 1% expertise this is 1.75%

For 60% avoidance, no expertise this is .96%
For 61% avoidance, no expertise, this is .8898%
For 60% avoidance, 1% expertise this is .896%

For 70/0 it's .405%
for 70/1 it's .378%
for 71/0 it's .3658%

So as you get into high levels of avoidance, avoidance stacking becomes better than expertise (marginally) in reducing the case of taking three hits in a row, one of which is hasted. That may be a better model to use in judging whether a tank gets gibbed; my experience is otherwise, but I haven't done extensive calculation in it. One thing to note is that with expertise it is possible to remove the chance entirely from the table; this isn't true for dodge (or not reasonably true). In any case the two are very similar in this situation. As you add more attacks into the equation without adding more hasted attacks avoidance will continue to improve thanks to the polynomial nature of it, but I don't think it's reasonable to worry about the situation where you take 4 or 5 hits in a row. That's not the likely case, and it's still very manageable by healing.

The assumption is that non-parry hasted hits do not kill tanks under most circumstances. Even a string of hits that aren't quick will not kill a tank; it'll make healers work harder, but it won't kill that tank. Heck, you say this yourself in the previous paragraph about tanking teron.

 

◊ Fellwraith

Teron is a bit different from most Sunwell encounters. There are very few encounters I can think of where 2 hits in rapid succession are going to kill a tank and most of them are in Sunwell. It takes a number of hits to get you to the point that you are going to get killed from a faster mob attack speed and avoiding any hits in that series makes it possible for healers to get you topped off so the parry flurry or subsequent attacks aren't a threat (this is particularly true if you have a druid healing you).

If you don't avoid void sentinel hits on M'uru, you're going to die - one and a half healers cannot keep up with every hit connecting. It doesn't matter if those hits are coming because of parry flurry or not; it's a simple question of healing throughput in a tightly tuned encounter. If you don't avoid one out of four hits during stomp, you are probably going to die (partially a moot point since Brutallus doesn't benefit from parries). If you don't avoid any hits during corrosion, you're going to be in serious trouble or pull too much healing off the raid. The burst in the Kalecgos encounter comes because you're stunned, it has nothing to do with your swings or lack of avoidance.


The argument is mostly academic however. You don't have the option to gem for expertise instead of avoidance. I'd assume you wouldn't ever equip something like shard of contempt over moroe's pocketwatch in a survival fight. Likewise, pendant of the titan's is clearly superior to the badge reward neck for survival. The reality is that you only have so many options for expertise gear and most of it already has a lot of avoidance on it.

 

kalbear

Right...but in all of those cases, expertise does nothing special anyway. Void Sentinels don't have a ton of parry. Felmyst won't kill you because of a parry-hasted attack either. Brutallus can't, and Kal won't. Clearly, expertise is a poor form of mitigation when parry-gibbing doesn't actually occur.

So yes, to be clear: expertise does not help when parry haste doesn't come into play or when the encounter requires something else.

This seems like a different question, honestly. Not 'what is the best way to avoid parry gib' but 'what actually kills tanks'. Expertise is better until you can stack very high levels of avoidance in avoiding parry gib. That doesn't mean it's better overall, but if your concern is about that facet, it's the best. If your concern is taking 4 hits in a row (regardless of the speed of those hits) that's a different question.

 

• Riot

What kills tanks?

Healers who aren't paying attention. Bosses that hit hard enough in a single swing are few and far between, or those huge hits that come at precise times - Stomp, Corrosion, Corrupting Strike.

Expertise is nice and all, but there are really no fights that I'd recommend stacking it. It's true - expertise will pretty much bring down hits to a predictable level. Unfortunately, bringing down damage to a predictable level of damage doesn't automatically make it manageable as well.

According to the LSAT examination, the opposite of hot is: A) Cold B) Not Hot

 

Jokal

A. To Kalbear, your math is incorrect due to the fact that simply multiplying by one factor of .15 for parry assumes that the tank only swings once during that whole cycle.


B. From a pure damage mitigation standpoint, Expertise does provide a rather useful benefit.

Random situation 1:
You swing 9 times every 6 seconds(SS/Rev/Dev/Dev cycle)
A boss swings 3 times every 6 seconds.

The average haste granted from parrying an attack is 24.2%. So given a 100 mob swing sample, 1% parry provides a .242% increase in damage. And given that a player swings 3 times as much as a mob, that number gets upped to .726%.

So 1% expertise or 4 skill or 15.7 rating decreases damage taken by .726%.

Compare that to dodge where 18.9 rating decreases damage taken by 1%.

How you value that nature of dodge vs the nature of expertise is a complete different topic, but there's the basic math.

Also, as that 9/3 figure approaches 1, the less valuable expertise becomes.

 

kalbear

The original factor was for a 2-second period. I didn't think that to be a particularly bad assumption given weapon swing timers. Factoring in multiple swings in that period would only increase the effectiveness of expertise.

Predictable damage is much easier to manage than unpredictable damage. That's kind of the point of EH, after all.

 

◊ Fellwraith

All the expertise in the world doesn't save you from random shit like this from happening (this was 2 weeks ago during a farming run):

22:08'47.353 Shadowmang's Swing parried by Teron Gorefiend
22:08'47.353 Shadowmang's Swing parried by Teron Gorefiend
22:08'47.451 Teron Gorefiend's Swing hits Mulack for 5803 Physical damage (756 blocked) <-- this was his normal swing
22:08'47.747 Utaki's Swing parried by Teron Gorefiend
22:08'47.747 Shadowmang's Sinister Strike parried by Teron Gorefiend
22:08'47.857 Teron Gorefiend's Swing hits Mulack for 4095 Physical damage (756 blocked) <-- Half a second swingspeed?
22:08'47.857 Flerix's Swing parried by Teron Gorefiend
22:08'48.169 Shadowdru's Rip parried by Teron Gorefiend
22:08'48.169 Flerix's Sinister Strike parried by Teron Gorefiend
22:08'48.169 Utaki's Swing parried by Teron Gorefiend
22:08'48.248 Teron Gorefiend's Swing hits Mulack for 9048 Physical damage (crushing) <--1 second later - his normal swing speed is about 2 seconds


He was turning to cast shadows of death and incinerate during those parries and proceeded to parry/shadowbolt gib me despite the 14% parry reduction and 60+% avoidance I was sporting. Was it bad luck? Yep. However, it's a very good real-world counterargument to the single-tank mathematical models people are throwing into this thread. Mobs turn to cast, channel spells (reducing their parry to 0), and you have to remember there are 24 other people in your raid (and their pets) who can be out of position periodically.

Expertise only goes so far. Your melee dps will never be parry-capped with expertise gear, it's a complete waste of their item budget 99% of the time. I think you have a false sense of security if you think expertise is going to completely reduce the burst on you.